Introduction
The Caribbean region weathered the global financial crisis relatively well, but the quality of bank assets gradually deteriorated during the subsequent economic recession, leaving many countries with elevated levels of problem loans. Notwithstanding significant heterogeneity across the region, the share of non-performing loans (NPLs) in total loans, which was relatively low before the global financial crisis, rose to more than 10 percent in 2016 across several Caribbean countries, peaking around 15 percent for many countries over the period 2007–16 (Figures 11.1 and 11.2).1 The increase in NPLs was more significant in tourism-dependent countries compared with commodity exporters. NPL ratios in many countries have been slow to decline from their elevated levels owing to structural and institutional impediments to resolution, as well as subdued or declining loan growth and sluggish economic activity, while in a number of countries, NPL ratios have started to fall from their peaks, owing to increased efforts undertaken by country authorities and banks to reduce impaired assets.

NPLs by Country
(Percent of total loans)
Sources: Country authorities; World Bank, World Development Indicators; and IMF staff estimates and calculations.Note: NPL = nonperforming loan. The latest available data for Suriname are for 2015.
NPLs by Country
(Percent of total loans)
Sources: Country authorities; World Bank, World Development Indicators; and IMF staff estimates and calculations.Note: NPL = nonperforming loan. The latest available data for Suriname are for 2015.NPLs by Country
(Percent of total loans)
Sources: Country authorities; World Bank, World Development Indicators; and IMF staff estimates and calculations.Note: NPL = nonperforming loan. The latest available data for Suriname are for 2015.
NPLs and Real GDP Growth
Sources: Country authorities; World Bank, World Development Indicators; and IMF staff estimates and calculations.Note: NPL = nonperforming loan.
NPLs and Real GDP Growth
Sources: Country authorities; World Bank, World Development Indicators; and IMF staff estimates and calculations.Note: NPL = nonperforming loan.NPLs and Real GDP Growth
Sources: Country authorities; World Bank, World Development Indicators; and IMF staff estimates and calculations.Note: NPL = nonperforming loan.The persistently high level of NPLs constrains credit and economic activity in the region and increases banks’ vulnerability to shocks. High NPLs have a negative impact on bank profitability, given that they require banks to raise provisioning and do not usually generate income (Aiyar and others 2015). Because Caribbean financial systems are dominated by banks, lack well-developed capital markets, and include a relatively small nonbank financial sector, persistent NPLs constrain the supply of credit to the private sector and hinder economic activity. High NPLs also increase banks’ vulnerability to shocks by reducing their profitability, tying up their capital, and raising funding costs. As such, banks with high NPLs present a potential risk to financial stability in the region.
This chapter analyzes the determinants and consequences of NPLs in the Caribbean, explores possible obstacles to their resolution, and suggests a strategy to move forward. It first takes stock of the extent of the NPL problem in the region and analyzes its characteristics and evolution. It then examines the determinants of NPLs and their importance as a driver of macro-financial feedback loops. Finally, it explores the main impediments to the resolution of NPLs in the region to draw policy implications for addressing the Caribbean NPL problem. To address these objectives, the paper uses three complementary approaches.
First, a panel vector autoregression (VAR) analysis uses country-level data to examine macro-financial links between NPLs, credit growth, and economic activity in the Caribbean economies. Empirical evidence confirms strong macro-financial links, with deteriorating asset quality lowering private credit growth and spilling over into the broader economy by depressing economic activity and resulting in higher unemployment. An improved macroeconomic environment, however, could enhance borrowers’ debt-service capacity and, other things being equal, result in lower NPLs.2
Second, dynamic panel regressions analyze the determinants of NPLs using both country data and detailed bank-level data. The results suggest that deteriorating asset quality can be attributed to both macroeconomic and bank-specific factors. More specifically, NPLs are affected by the business cycle—low economic growth, including in advanced economies, leads to more problem loans, particularly in tourism-dependent economies. After controlling for endogeneity between NPLs and bank fundamentals, results from similar regressions with a novel bank-level data set also suggest that banks with weaker fundamentals (lower profitability, capital adequacy, and efficiency) also tend to suffer from weaker asset quality.
Third, a survey of regional authorities and commercial banks explores obstacles to NPL resolution in the Caribbean. Consistent with the empirical evidence, survey responses highlight the importance of low growth and high unemployment, as well as weaknesses in regional real estate markets, as determinants of high and persistent NPLs. In addition, the results point to interrelated and mutually reinforcing structural obstacles to NPL resolution. Specifically, the results suggest that the lack of markets for distressed assets, gaps in information systems, and deficiencies in the legal system and insolvency and debt-enforcement regimes are key obstacles to NPL resolution.
The analysis suggests that problem loans should be tackled on multiple fronts. In addition to pursuing policies aimed at securing strong and sustainable growth, as well as micro- and macroprudential policies to limit excessive risk taking during loan origination, the authorities should develop targeted strategies for addressing structural obstacles to the resolution of NPLs. Policies should aim to address information gaps and eliminate impediments to information sharing to facilitate valuation of collateral and development of markets for distressed assets; address deficiencies in insolvency and debt-enforcement frameworks to accelerate and maximize recovery, including through strengthening judicial systems; and establish conditions for nonbanks specialized in servicing problem loans to facilitate collection and disposal of distressed assets. Requiring provisioning independent of collateral values could ensure swift write-off. Consideration could also be given to establishing a pan-Caribbean market for problem loans that would be open to external investors.
Taking Stock: How Serious is the Problem?
NPLs have risen sharply across the Caribbean following the global financial crisis. Since the crisis, banking systems in many Caribbean countries have been plagued by high ratios of NPLs, which, except in Jamaica and Trinidad and Tobago, are well above prudential guidelines. The average NPL ratio across the Caribbean is also high compared with most other regions in the world. Country-level data for the past 20 years point to a significant deterioration of bank asset quality after the global financial crisis, unlike in Latin America, where the impact was more subdued (Figure 11.3). This outcome could be attributed to stronger macroeconomic fundamentals, on average, in Latin America compared with the Caribbean; greater export diversification; and relatively weaker links to advanced economies, especially the United States. Moreover, subdued bank profitability in the Caribbean has limited the ability of many banks to adequately provision for problem loans, with provisioning to NPLs low by international comparison—averaging at about 60 percent, though with significant variation within the region.3

Cross-Country NPLs and Provisions
Sources: Country authorities; and IMF, Financial Soundness Indicators database.Note: CIS = Commonwealth of Independent States; MENA = Middle East and North Africa; NPL = nonperforming loan.
Cross-Country NPLs and Provisions
Sources: Country authorities; and IMF, Financial Soundness Indicators database.Note: CIS = Commonwealth of Independent States; MENA = Middle East and North Africa; NPL = nonperforming loan.Cross-Country NPLs and Provisions
Sources: Country authorities; and IMF, Financial Soundness Indicators database.Note: CIS = Commonwealth of Independent States; MENA = Middle East and North Africa; NPL = nonperforming loan.The high NPLs in the Caribbean are, in large part, a legacy of the global financial crisis, but also reflect some structural problems. Before the crisis, credit growth was strong, spurred by economic activity in the tourism industry and related construction, as well as by favorable global commodity prices for the commodity-exporting Caribbean countries. The credit expansion led to higher private and public sector debt, increasing borrowers’ vulnerability to shocks and banks’ exposure to credit risk. The crisis was transmitted to the region primarily through lower demand for the region’s tourism services. Lower tourism arrivals had a significant negative impact on tourism-dependent industries and related construction, resulting in a sharp increase in unemployment and loss of income for households, which, in turn, impaired their ability to service their loans. As a result, the increase in NPLs was also concentrated in the personal, construction, and tourism sectors (Figure 11.4). In many countries, NPL ratios in these sectors remained elevated subsequent to the crisis, consistent with the slow pace of economic recovery experienced by much of the region. Persistently high NPLs, however, also reflected banks’ slow pace of restructuring, sale of NPLs, and write-offs, even after a gradual pickup in economic activity. In most economies, NPL ratios remain above their precrisis levels. In several countries, including Antigua and Barbuda, Belize, and Jamaica, NPL ratios have been falling steadily since 2012 toward their precrisis levels.


Although remaining high, there is considerable heterogeneity in the ratio of NPLs across banks in the region. As of 2015, the NPL ratio varied from about 40 percent of total loans to close to zero. The distribution of NPLs has widened, with more banks experiencing higher NPL ratios after the crisis (Figure 11.5). Domestic (that is, locally owned) banks have higher average NPL ratios than foreign-owned banks, but provisioning ratios, which were much lower in domestic banks before the crisis, are now about the same as foreign-owned banks’ provisioning.

Bank-Level NPLs and Provisioning
Sources: Country authorities; and IMF staff calculations.Note: NPL = nonperforming loan.
Bank-Level NPLs and Provisioning
Sources: Country authorities; and IMF staff calculations.Note: NPL = nonperforming loan.Bank-Level NPLs and Provisioning
Sources: Country authorities; and IMF staff calculations.Note: NPL = nonperforming loan.Macro-Financial Implications of High NPLs
High levels of NPLs affect bank lending and may result in adverse macro-financial feedback loops. High NPLs typically reduce the supply of credit, including by reducing bank profitability through higher provisioning, by tying up capital because of higher risk weights on impaired assets, and by raising banks’ funding costs because of lower expected revenue streams and investors’ heightened risk perceptions (Aiyar and others). Following the global financial crisis, Caribbean banks tightened their lending standards as they focused on cleaning up their balance sheets, reducing extension of credit and downsizing and consolidating their operations to compensate for reduced income flows from NPLs. Reduced credit supply, in turn, contributed to weaker economic activity, with adverse implications for NPLs. Country-level data indicate that NPL ratios have indeed been negatively correlated with bank profitability and private sector credit growth, suggesting that banks with higher NPL ratios have lower profitability and lend less (Figures 11.6 and 11.7), while private credit growth is positively correlated with economic activity, suggesting that high NPLs would be associated with subdued growth.

NPLs and Bank Profitability
(Percent)
Sources: Country authorities; World Bank, World Development Indicators; IMF, World Economic Outlook; and IMF staff estimates and calculations.Note: NPL = nonperforming loan.
NPLs and Bank Profitability
(Percent)
Sources: Country authorities; World Bank, World Development Indicators; IMF, World Economic Outlook; and IMF staff estimates and calculations.Note: NPL = nonperforming loan.NPLs and Bank Profitability
(Percent)
Sources: Country authorities; World Bank, World Development Indicators; IMF, World Economic Outlook; and IMF staff estimates and calculations.Note: NPL = nonperforming loan.
NPLs, Private Credit, and Economic Activity
Sources: Country authorities; World Bank, World Development Indicators; IMF, World Economic Outlook; and IMF staff estimates and calculations.Note: NPL = nonperforming loan.
NPLs, Private Credit, and Economic Activity
Sources: Country authorities; World Bank, World Development Indicators; IMF, World Economic Outlook; and IMF staff estimates and calculations.Note: NPL = nonperforming loan.NPLs, Private Credit, and Economic Activity
Sources: Country authorities; World Bank, World Development Indicators; IMF, World Economic Outlook; and IMF staff estimates and calculations.Note: NPL = nonperforming loan.A panel VAR model is estimated to more formally assess feedback effects between asset quality in the banking sector and the real economy. The approach allows examining the interactions among variables, including the duration and magnitude of the effect. The model includes five endogenous variables (change in the ratio of NPLs to total loans, growth of credit to the private sector, change in the unemployment rate, real GDP growth, and consumer price inflation), a vector of exogenous variables (change in the U.S. unemployment rate and a dummy variable for natural disasters), and country fixed effects.4 Because the unemployment rate is unavailable for five out of 13 countries in the sample, the baseline model is estimated for eight Caribbean economies, using an unbalanced panel of annual observations for the period 1997–2015, with the major macroeconomic indicators unavailable at a higher frequency.5 A model excluding unemployment is also estimated for the full sample. Results are broadly robust to alternative specifications.
The results of the analysis point to the presence of strong macro-financial links in Caribbean economies. A shock to the NPL ratio has a statistically significant impact on private sector credit growth, economic activity (Figure 11.8), and the unemployment rate. Specifically, worsening asset quality has persistent negative implications for credit growth, which can last up to five years. The persistent nature of credit cycles also suggests that a shock to credit growth continues for several years. Lower private credit growth, in turn, reduces economic activity and results in an increase in unemployment (Annex 11.1). At the same time, macroeconomic performance also has a significant effect on asset quality: stronger economic performance leads to a statistically significant decline in the NPL ratio (Figure 11.9). Variance decompositions (not shown) indicate that over a five-year horizon, about 22 percent of the variance in the NPL ratios in the productive sectors (including construction, tourism, agriculture, and manufacturing) is driven by shocks to real GDP growth, suggesting that macroeconomic performance is an important determinant of the NPL ratio. Meanwhile, the impact of the NPL ratio on economic performance is smaller, with shocks to the NPL ratio explaining about 7 percent of the variance in real GDP growth.

PVAR Results: Shock to NPLs
Source: IMF staff estimates and calculations.Note: Impulse response functions to shocks of one standard deviation. Dashed lines show errors of 10 percent generated by Monte Carlo with 300 simulations. NPL = nonperforming loan; PVAR = panel vector autoregression.
PVAR Results: Shock to NPLs
Source: IMF staff estimates and calculations.Note: Impulse response functions to shocks of one standard deviation. Dashed lines show errors of 10 percent generated by Monte Carlo with 300 simulations. NPL = nonperforming loan; PVAR = panel vector autoregression.PVAR Results: Shock to NPLs
Source: IMF staff estimates and calculations.Note: Impulse response functions to shocks of one standard deviation. Dashed lines show errors of 10 percent generated by Monte Carlo with 300 simulations. NPL = nonperforming loan; PVAR = panel vector autoregression.
PVAR Results: Shock to Real GDP Growth
Source: IMF staff estimates and calculations.Note: Impulse response functions to shocks of one standard deviation. Dashed lines show errors of 10 percent generated by Monte Carlo with 300 simulations. NPL = nonperforming loan; PVAR = panel vector autoregression.
PVAR Results: Shock to Real GDP Growth
Source: IMF staff estimates and calculations.Note: Impulse response functions to shocks of one standard deviation. Dashed lines show errors of 10 percent generated by Monte Carlo with 300 simulations. NPL = nonperforming loan; PVAR = panel vector autoregression.PVAR Results: Shock to Real GDP Growth
Source: IMF staff estimates and calculations.Note: Impulse response functions to shocks of one standard deviation. Dashed lines show errors of 10 percent generated by Monte Carlo with 300 simulations. NPL = nonperforming loan; PVAR = panel vector autoregression.The results are consistent with what is suggested by a simple correlation analysis. Simple correlations of key macro-financial variables indicate that the NPL ratio is negatively correlated with inflation, foreign direct investment, real GDP, and credit growth, and positively correlated with the unemployment rate (Table 11.1).
Correlation across Macro-Financial Variables
(Annual frequency)
Correlation across Macro-Financial Variables
(Annual frequency)
NPL | Loan | FDI | U | GDP | CPI | ||
---|---|---|---|---|---|---|---|
NPL Ratio Change | NPL | 1 | |||||
Loan Growth | Loan | −0.1377* | 1 | ||||
FDI Growth | FDI | −0.0057 | −0.1317* | 1 | |||
Change in Unemployment Rate | U | 0.2895* | −0.2525* | −0.1113 | 1 | ||
Real GDP Growth | GDP | −0.1225* | 0.2524* | 0.0159 | −0.3779* | 1 | |
Inflation | CPI | −0.0309 | 0.4015* | −0.0152 | 0.0101 | −0.1289* | 1 |
Correlation across Macro-Financial Variables
(Annual frequency)
NPL | Loan | FDI | U | GDP | CPI | ||
---|---|---|---|---|---|---|---|
NPL Ratio Change | NPL | 1 | |||||
Loan Growth | Loan | −0.1377* | 1 | ||||
FDI Growth | FDI | −0.0057 | −0.1317* | 1 | |||
Change in Unemployment Rate | U | 0.2895* | −0.2525* | −0.1113 | 1 | ||
Real GDP Growth | GDP | −0.1225* | 0.2524* | 0.0159 | −0.3779* | 1 | |
Inflation | CPI | −0.0309 | 0.4015* | −0.0152 | 0.0101 | −0.1289* | 1 |
Determinants of NPLs
Country-level and bank-level panel regression analyses are used to more systematically assess the determinants of NPLs in the Caribbean. A dynamic panel regression is estimated, first to analyze the determinants of NPLs at an annual frequency, focusing on macroeconomic determinants of NPLs, and subsequently to analyze the bank-specific determinants of NPLs at a higher frequency. The dynamic panel regression specification for the annual country-level regressions is given by
where NPLsj,t denotes the logit transformation of the NPL ratio for country j at time t. The dependent variable is explained by its lag (NPLsj,t-1), global (Globalt) and country-specific (Countryj,t) macroeconomic variables, and country fixed effects (uj).The term k is the number of annual lags. The model is estimated with an unbalanced panel of 13 Caribbean countries.6 The same baseline model is estimated using bank-level data for bank i in country j at time t with quarterly data. In addition to the variables considered in equation (11.1), the bank-level model includes detailed bank-level (Banki,j,t) variables capturing credit risk (including profitability, capital adequacy, measures of efficiency the expense-to-assets ratio, bank size, and credit concentration, as well as bank-level fixed effects.7 The bank-level model is estimated with an unbalanced panel of 71 banks across 13 Caribbean countries.
Country-level regressions also suggest that economic activity is an important determinant of banks’ asset quality. Asset quality deteriorates as a function of low real GDP growth, both at home and in advanced economies (Table 11.2). Spillovers from global macroeconomic developments are particularly strong, with the coefficient on real GDP growth in advanced economies larger and statistically more significant than that on domestic growth, consistent with the high degree of openness of economies in the region. Asset quality also deteriorates as a function of higher lending rates that reduce borrowers’ capacity to service debt. Reduced credit growth also results in a higher NPL ratio, suggesting that NPLs are reduced at a slower rate than the fall in total loans. Estimating the regression separately for the periods before and after 2009 suggests that the global financial crisis explains much of the variation in asset quality, capturing the impact of macroeconomic developments in the postcrisis period.8
Macroeconomic Determinants of NPLs
Macroeconomic Determinants of NPLs
Full Sample | Pre-2009 | Post-2009 | |||||
---|---|---|---|---|---|---|---|
(1) | (2) | (3) | (4) | (5) | (6) | (7) | |
NPL (t-1) | 1.060*** (0.0935) | 0.891*** (0.0803) | 0.895*** (0.0665) | 0.896*** (0.0655) | 0.894*** (0.0665) | 0.677*** (0.0727) | 0.692*** (0.188) |
Real GDP Growth (t-1) | −0.0195* (0.0112) | −0.00906 (0.0122) | −0.00961 (0.00756) | −0.00987 (0.00768) | −0.00973 (0.00775) | −0.0111* (0.00648) | −0.00551 (0.0190) |
Advanced Economies Real GDP Growth (t-1) | −0.0242** | −0.0170** | −0.0236*** | −0.0241*** | −0.0233*** | −0.0355 | −0.0112 |
Credit Growth (t) | (0.0102) | (0.00813) −0.0128*** | (0.00710) −0.0121*** | (0.00746) −0.0124*** | (0.00711) −0.0122*** | (0.0586) −0.0101*** | (0.0121) −0.00994 |
Lending Rate (t-1) | (0.00327) | (0.00267) 0.0225*** | (0.00309) 0.0224*** | (0.00285) 0.0237*** | (0.00221) 0.0133 | (0.00691) −0.0591 | |
Real Effective Exchange Rate Growth (t-1) | (0.00726) | (0.00702) −0.00180 | (0.00704) | (0.0164) | (0.0731) | ||
(0.00389) | |||||||
Natural Disaster | 0.00461 (0.0528) | ||||||
Constant | 0.279 (0.241) | −0.0992 (0.213) | −0.398** (0.174) | −0.387** (0.159) | −0.416** (0.170) | −0.905** (0.352) | 0.0160 (0.602) |
Observations | 229 | 223 | 162 | 162 | 162 | 78 | 84 |
Number of Countries | 13 | 13 | 12 | 12 | 12 | 11 | 12 |
Number of | 8 | 11 | 14 | 15 | 15 | 14 | 14 |
Instruments | |||||||
AR(1) | 0.0144 | 0.0149 | 0.0318 | 0.0313 | 0.0313 | 0.158 | 0.108 |
AR(2) | 0.426 | 0.462 | 0.333 | 0.327 | 0.337 | 0.385 | 0.226 |
Hansen | 0.329 | 0.257 | 0.289 | 0.546 | 0.571 | 0.615 | 0.510 |
Macroeconomic Determinants of NPLs
Full Sample | Pre-2009 | Post-2009 | |||||
---|---|---|---|---|---|---|---|
(1) | (2) | (3) | (4) | (5) | (6) | (7) | |
NPL (t-1) | 1.060*** (0.0935) | 0.891*** (0.0803) | 0.895*** (0.0665) | 0.896*** (0.0655) | 0.894*** (0.0665) | 0.677*** (0.0727) | 0.692*** (0.188) |
Real GDP Growth (t-1) | −0.0195* (0.0112) | −0.00906 (0.0122) | −0.00961 (0.00756) | −0.00987 (0.00768) | −0.00973 (0.00775) | −0.0111* (0.00648) | −0.00551 (0.0190) |
Advanced Economies Real GDP Growth (t-1) | −0.0242** | −0.0170** | −0.0236*** | −0.0241*** | −0.0233*** | −0.0355 | −0.0112 |
Credit Growth (t) | (0.0102) | (0.00813) −0.0128*** | (0.00710) −0.0121*** | (0.00746) −0.0124*** | (0.00711) −0.0122*** | (0.0586) −0.0101*** | (0.0121) −0.00994 |
Lending Rate (t-1) | (0.00327) | (0.00267) 0.0225*** | (0.00309) 0.0224*** | (0.00285) 0.0237*** | (0.00221) 0.0133 | (0.00691) −0.0591 | |
Real Effective Exchange Rate Growth (t-1) | (0.00726) | (0.00702) −0.00180 | (0.00704) | (0.0164) | (0.0731) | ||
(0.00389) | |||||||
Natural Disaster | 0.00461 (0.0528) | ||||||
Constant | 0.279 (0.241) | −0.0992 (0.213) | −0.398** (0.174) | −0.387** (0.159) | −0.416** (0.170) | −0.905** (0.352) | 0.0160 (0.602) |
Observations | 229 | 223 | 162 | 162 | 162 | 78 | 84 |
Number of Countries | 13 | 13 | 12 | 12 | 12 | 11 | 12 |
Number of | 8 | 11 | 14 | 15 | 15 | 14 | 14 |
Instruments | |||||||
AR(1) | 0.0144 | 0.0149 | 0.0318 | 0.0313 | 0.0313 | 0.158 | 0.108 |
AR(2) | 0.426 | 0.462 | 0.333 | 0.327 | 0.337 | 0.385 | 0.226 |
Hansen | 0.329 | 0.257 | 0.289 | 0.546 | 0.571 | 0.615 | 0.510 |
The bank-level analysis shows that both macroeconomic and bank-specific factors are important determinants of NPLs (Table 11.3). NPLs are persistent (suggested by the coefficient of lagged NPLs), pointing to difficulties in resolving problem loans. Deterioration in advanced economy growth worsens asset quality. Tourism arrivals, included as a proxy for domestic economic conditions, given that real GDP is unavailable for most countries in the region on a quarterly basis, are, by contrast, statistically insignificant. The effect of tourism arrivals is likely captured by advanced economy growth, which is an important determinant of tourism arrivals to the region. Bank-specific factors are also important determinants of credit risk and asset quality. More profitable, more capitalized, and more efficient banks (as measured by expense-to-asset and income-to-expense ratios) tend to have better asset quality. Exposure to credit risk associated with higher lending to the household sector and higher foreign currency lending does not seem to be a significant determinant of NPLs at the bank level. A priori, one would expect the foreign banks—primarily subsidiaries and branches of Canadian banks with global operations—to have stronger risk management practices and greater capacity to dispose of impaired assets. However, the results suggest that asset quality is not affected by whether banks are foreign or domestically owned.9 The significant dispersion in NPL ratios across banks likely explains this result.10
Bank-Level Determinants of NPLs: Bank Performance
Bank-Level Determinants of NPLs: Bank Performance
Baseline | Profitability | Capital Adequacy | Bank Efficiency | Credit Concentration | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
(1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) | (12) | |
NPL(t-l) | 0.631*** (0.0630) | 0.624*** (0.0685) | 0.608*** (0.0723) | 0.614*** (0.0800) | 0.591*** (0.0804) | 0.591*** (0.0772) | 0.789*** (0.0940) | 0.783*** (0.0803) | 0.789*** (0.0873) | 0.579*** (0.0935) | 0.641*** (0.0671) | 0.733*** (0.0894) |
NPL (t-2) | 0.190*** (0.0569) | 0.151*** (0.0511) | 0.144*** (0.0525) | 0.128** (0.0637) | 0.144*** (0.0535) | 0.124** (0.0559) | 0.0284 (0.0669) | 0.0351 (0.0631) | 0.0329 (0.0649) | 0.146*** (0.0545) | 0.146*** (0.0537) | 0.183*** (0.0537) |
Advanced Economies Real GDP Growth (t-1) | −0.00704** (0.00330) | −0.00597* (0.00309) | −0.00674** (0.00337) | −0.00480 (0.00346) | −0.00533 (0.00351) | −0.00569 (0.00380) | −0.00285 (0.00407) | −0.00286 (0.00307) | −0.00342 (0.00344) | −0.00783 (0.00602) | −0.00473 (0.00337) | −0.0119*** (0.00415) |
Credit Growth (t) | −0.00452*** (0.00100) | −0.00429*** (0.000903) | −0.00417*** (0.000933) | −0.00405** (0.00180) | −0.00376*** (0.00100) | −0.00496*** (0.00117) | −0.00351** (0.00142) | −0.00353*** (0.00129) | −0.00349** (0.00136) | −0.00461*** (0.00144) | −0.00478*** (0.00106) | −0.00284*** (0.000961) |
Return on Assets (t) | −0.0122* (0.00641) | −0.0144* (0.00756) | −0.0150* (0.00785) | −0.00278 (0.00636) | 0.000681 (0.00429) | −0.000286 (0.00457) | −0.0157* (0.00833) | −0.0106 (0.00866) | −0.0119 (0.00749) | |||
Return on Equity (t) | −0.000291 (0.000219) | |||||||||||
Net Interest Margin (t) | −0.0810* (0.0428) | |||||||||||
Capital Adequacy Ratio (t) | −0.00582* (0.00331) | −0.00396 (0.00394) | 0.00779 (0.00874) | 0.00431 (0.00686) | 0.00786 (0.00772) | −0.00530 (0.00359) | ||||||
Loan-to-Deposit Ratio (t-1) | −0.00618** (0.00307) | |||||||||||
Expense-to-Assets Ratio (t-1) | 0.0906** (0.0376) | 0.0263 (0.0512) | ||||||||||
Income-to-Expenses Ratio (t) | −0.202*** (0.0697) | −0.181* (0.0925) | ||||||||||
Total Assets (In) (t) | 0.0459 (0.105) | |||||||||||
Loans to Households (percent of total) (t) | −0.00341 (0.00304) | |||||||||||
Foreign Currency Loans (percent of total) (t) | −2.31e-06 (3.56e-06) | |||||||||||
Constant | −0.427 (0.273) | −0.533** (0.269) | −0.598** (0.285) | −0.545 (0.345) | −0.577* (0.304) | −0.195 (0.183) | −0.746** (0.352) | −0.238 (0.181) | −0.374 (0.230) | −1.133 (1.508) | −0.360 (0.303) | −0.145 (0.210) |
Observations | 3,709 | 3,508 | 3,363 | 3,286 | 3,063 | 3,063 | 766 | 766 | 766 | 3,063 | 3,283 | 948 |
Number of Banks | 71 | 71 | 66 | 62 | 62 | 62 | 22 | 22 | 22 | 62 | 62 | 28 |
Number of Instruments | 10 | 13 | 13 | 10 | 16 | 19 | 19 | 19 | 22 | 19 | 16 | 16 |
AR(1) | 3.58e-07 | 5.41 e-07 | 7.60e-07 | 2.05e-06 | 2.79e-06 | 2.35e-06 | 0.00985 | 0.00917 | 0.00914 | 2.92e-06 | 1.47e-06 | 0.00105 |
AR(2) | 0.437 | 0.870 | 0.687 | 0.749 | 0.634 | 0.232 | 0.771 | 0.729 | 0.771 | 0.751 | 0.921 | 0.551 |
Hansen | 0.485 | 0.446 | 0.756 | 0.520 | 0.624 | 0.484 | 0.836 | 0.564 | 0.460 | 0.650 | 0.282 | 0.913 |
Bank-Level Determinants of NPLs: Bank Performance
Baseline | Profitability | Capital Adequacy | Bank Efficiency | Credit Concentration | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
(1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) | (12) | |
NPL(t-l) | 0.631*** (0.0630) | 0.624*** (0.0685) | 0.608*** (0.0723) | 0.614*** (0.0800) | 0.591*** (0.0804) | 0.591*** (0.0772) | 0.789*** (0.0940) | 0.783*** (0.0803) | 0.789*** (0.0873) | 0.579*** (0.0935) | 0.641*** (0.0671) | 0.733*** (0.0894) |
NPL (t-2) | 0.190*** (0.0569) | 0.151*** (0.0511) | 0.144*** (0.0525) | 0.128** (0.0637) | 0.144*** (0.0535) | 0.124** (0.0559) | 0.0284 (0.0669) | 0.0351 (0.0631) | 0.0329 (0.0649) | 0.146*** (0.0545) | 0.146*** (0.0537) | 0.183*** (0.0537) |
Advanced Economies Real GDP Growth (t-1) | −0.00704** (0.00330) | −0.00597* (0.00309) | −0.00674** (0.00337) | −0.00480 (0.00346) | −0.00533 (0.00351) | −0.00569 (0.00380) | −0.00285 (0.00407) | −0.00286 (0.00307) | −0.00342 (0.00344) | −0.00783 (0.00602) | −0.00473 (0.00337) | −0.0119*** (0.00415) |
Credit Growth (t) | −0.00452*** (0.00100) | −0.00429*** (0.000903) | −0.00417*** (0.000933) | −0.00405** (0.00180) | −0.00376*** (0.00100) | −0.00496*** (0.00117) | −0.00351** (0.00142) | −0.00353*** (0.00129) | −0.00349** (0.00136) | −0.00461*** (0.00144) | −0.00478*** (0.00106) | −0.00284*** (0.000961) |
Return on Assets (t) | −0.0122* (0.00641) | −0.0144* (0.00756) | −0.0150* (0.00785) | −0.00278 (0.00636) | 0.000681 (0.00429) | −0.000286 (0.00457) | −0.0157* (0.00833) | −0.0106 (0.00866) | −0.0119 (0.00749) | |||
Return on Equity (t) | −0.000291 (0.000219) | |||||||||||
Net Interest Margin (t) | −0.0810* (0.0428) | |||||||||||
Capital Adequacy Ratio (t) | −0.00582* (0.00331) | −0.00396 (0.00394) | 0.00779 (0.00874) | 0.00431 (0.00686) | 0.00786 (0.00772) | −0.00530 (0.00359) | ||||||
Loan-to-Deposit Ratio (t-1) | −0.00618** (0.00307) | |||||||||||
Expense-to-Assets Ratio (t-1) | 0.0906** (0.0376) | 0.0263 (0.0512) | ||||||||||
Income-to-Expenses Ratio (t) | −0.202*** (0.0697) | −0.181* (0.0925) | ||||||||||
Total Assets (In) (t) | 0.0459 (0.105) | |||||||||||
Loans to Households (percent of total) (t) | −0.00341 (0.00304) | |||||||||||
Foreign Currency Loans (percent of total) (t) | −2.31e-06 (3.56e-06) | |||||||||||
Constant | −0.427 (0.273) | −0.533** (0.269) | −0.598** (0.285) | −0.545 (0.345) | −0.577* (0.304) | −0.195 (0.183) | −0.746** (0.352) | −0.238 (0.181) | −0.374 (0.230) | −1.133 (1.508) | −0.360 (0.303) | −0.145 (0.210) |
Observations | 3,709 | 3,508 | 3,363 | 3,286 | 3,063 | 3,063 | 766 | 766 | 766 | 3,063 | 3,283 | 948 |
Number of Banks | 71 | 71 | 66 | 62 | 62 | 62 | 22 | 22 | 22 | 62 | 62 | 28 |
Number of Instruments | 10 | 13 | 13 | 10 | 16 | 19 | 19 | 19 | 22 | 19 | 16 | 16 |
AR(1) | 3.58e-07 | 5.41 e-07 | 7.60e-07 | 2.05e-06 | 2.79e-06 | 2.35e-06 | 0.00985 | 0.00917 | 0.00914 | 2.92e-06 | 1.47e-06 | 0.00105 |
AR(2) | 0.437 | 0.870 | 0.687 | 0.749 | 0.634 | 0.232 | 0.771 | 0.729 | 0.771 | 0.751 | 0.921 | 0.551 |
Hansen | 0.485 | 0.446 | 0.756 | 0.520 | 0.624 | 0.484 | 0.836 | 0.564 | 0.460 | 0.650 | 0.282 | 0.913 |
Obstacles to NPL Resolution
Although the high level of NPLs in the Caribbean has been, in large part, a legacy of the global financial crisis, their persistence owes much to the weak economic recovery in the region, as well as to structural obstacles to resolution. The pace of loan restructuring, NPL sales, and write-offs has been slow, leaving large amounts of impaired assets on bank balance sheets. Results from a survey of country authorities and banks operating in the region highlight some common elements undermining NPL resolution. This section summarizes the results of the survey, which asked banks and authorities to (1) identify the key obstacles to NPL resolution; (2) provide detailed information on structural impediments grouped into five broad areas: supervisory and prudential, insolvency and debt enforcement, informational systems, market for distressed debt, and tax regime; and (3) identify the key measures undertaken by the authorities and banks in recent years toward resolution of NPLs (Annex 11.1).11 Responses were received from the authorities of the Eastern Caribbean Central Bank, The Bahamas, Barbados, Belize, Guyana, Jamaica, Montserrat, St. Kitts and Nevis, Suriname, and Trinidad and Tobago, and 39 banks representing 40 percent of all banks in the region.
Survey responses from the authorities and banks highlight macroeconomic conditions, deficiencies in the legal process, and difficulties with collateral as the key obstacles to NPL resolution (Figure 11.10). Most banks identified macroeconomic conditions, such as slow economic growth, high unemployment, and loss of clients’ incomes, as among the most severe obstacles to resolving NPLs. Many banks also pointed to deficiencies in the legal process (including, for example, costly and protracted insolvency procedures, or the absence of specialized courts or judges), difficulties in valuing and realizing collateral, and depressed real estate markets as major obstacles. The authorities, for their part, highlighted deficiencies in the legal process and banks’ underwriting and monitoring practices and inability or limited financial capacity in certain banks to recognize loan impairment and provision against it, as well as poor markets for distressed debt and collateral, as top obstacles, followed by macroeconomic factors.

Top Obstacles to NPL Resolution in the Caribbean
Source: IMF staff survey of national authorities and banks.Note: NPL = nonperforming loan. The figure summarizes responses to a question asking respondents to name top three obstacles to NPL resolution. Panel 2 combines responses from The Bahamas, Barbados, Belize, Guyana, Jamaica, Montserrat, St. Kitts and Nevis, and Trinidad and Tobago.
Top Obstacles to NPL Resolution in the Caribbean
Source: IMF staff survey of national authorities and banks.Note: NPL = nonperforming loan. The figure summarizes responses to a question asking respondents to name top three obstacles to NPL resolution. Panel 2 combines responses from The Bahamas, Barbados, Belize, Guyana, Jamaica, Montserrat, St. Kitts and Nevis, and Trinidad and Tobago.Top Obstacles to NPL Resolution in the Caribbean
Source: IMF staff survey of national authorities and banks.Note: NPL = nonperforming loan. The figure summarizes responses to a question asking respondents to name top three obstacles to NPL resolution. Panel 2 combines responses from The Bahamas, Barbados, Belize, Guyana, Jamaica, Montserrat, St. Kitts and Nevis, and Trinidad and Tobago.Country-by-country survey responses highlight several key areas of concern in resolving NPLs. Among these, survey respondents ranked the absence of a market for distressed assets as the most challenging area for NPL resolution across the Caribbean (Figure 11.11). As in many European countries with high NPLs (Aiyar and others 2015), the market for trading distressed assets is largely nonexistent in the Caribbean despite the absence of explicit restrictions (Figure 11.12). The survey responses also suggest concerns about gaps in the information framework, deficiencies in insolvency and debt-enforcement regimes (including the legal system, institutional weaknesses, and collateral enforcement related challenges), and weaknesses in prudential and supervisory frameworks. The tax systems in the Caribbean were perceived as posing less concern than other broad areas. Many elements are common across countries; however, detailed responses show significant variation across countries and banks.

The Caribbean: Survey-Based Scores on Obstacles to NPL Resolution by Country
Source: IMF staff survey of national authorities and banks.Note: NPL = nonperforming loan. The heat map shows maximum scores from country and bank surveys. The institutional obstacle scores are compiled from individual responses to these surveys. The surveys asked the respondents to indicate their level of concern regarding specific obstacles within five broad areas (such as NPL markets, information obstacles, and so forth), and provide information regarding these areas. Banks’ responses within the same jurisdictions were aggregated by averaging. The overall obstacle level is a simple average of obstacle scores in each of the five areas. Values 2.0 appearing a red box are marginally higher than 2.
The Caribbean: Survey-Based Scores on Obstacles to NPL Resolution by Country
Source: IMF staff survey of national authorities and banks.Note: NPL = nonperforming loan. The heat map shows maximum scores from country and bank surveys. The institutional obstacle scores are compiled from individual responses to these surveys. The surveys asked the respondents to indicate their level of concern regarding specific obstacles within five broad areas (such as NPL markets, information obstacles, and so forth), and provide information regarding these areas. Banks’ responses within the same jurisdictions were aggregated by averaging. The overall obstacle level is a simple average of obstacle scores in each of the five areas. Values 2.0 appearing a red box are marginally higher than 2.The Caribbean: Survey-Based Scores on Obstacles to NPL Resolution by Country
Source: IMF staff survey of national authorities and banks.Note: NPL = nonperforming loan. The heat map shows maximum scores from country and bank surveys. The institutional obstacle scores are compiled from individual responses to these surveys. The surveys asked the respondents to indicate their level of concern regarding specific obstacles within five broad areas (such as NPL markets, information obstacles, and so forth), and provide information regarding these areas. Banks’ responses within the same jurisdictions were aggregated by averaging. The overall obstacle level is a simple average of obstacle scores in each of the five areas. Values 2.0 appearing a red box are marginally higher than 2.
International Comparison
(Survey-based scores of obstacles to NPL resolution)
Sources: Aiyar and others (2015); and surveys of banks and the national authorities in the Caribbean.Note: NPL = nonperforming loan.1 Countries with NPL ratio exceeding 10 percent during 2005–14 (Aiyar and others 2015).
International Comparison
(Survey-based scores of obstacles to NPL resolution)
Sources: Aiyar and others (2015); and surveys of banks and the national authorities in the Caribbean.Note: NPL = nonperforming loan.1 Countries with NPL ratio exceeding 10 percent during 2005–14 (Aiyar and others 2015).International Comparison
(Survey-based scores of obstacles to NPL resolution)
Sources: Aiyar and others (2015); and surveys of banks and the national authorities in the Caribbean.Note: NPL = nonperforming loan.1 Countries with NPL ratio exceeding 10 percent during 2005–14 (Aiyar and others 2015).Prudential and Supervisory Framework and NPL Management
Strong supervision and conservative bank practices for loss recognition and provisioning increase incentives to address NPLs in a timely manner. Supervisory authorities in the region generally paid close attention to problem loans, tightening regulations in recent years. In most countries, supervisors have undertaken a thematic review of banks’ NPL management capacity, issued formal guidelines to banks on NPL management practices, and issued regulations regarding provisioning (or communicated regulatory expectations with regard to provisions under International Financial Reporting Standards). In some countries, supervisors provided additional incentives for write-offs, such as increased capital charges or time limits for carrying NPLs on balance sheets; performed assessments of collateral valuation practices; and subjected banks to granular asset quality reviews. In some countries, supervisors have strengthened prudential rules for loan classification and provisioning.
Notwithstanding these efforts, further work is needed in several areas. In most countries, no licensing and regulatory regime is in place to enable nonbanks to own or manage NPLs, which hampers the development of the NPL market. Many countries have no requirement to apply a real estate valuation standard, which contributes to uncertainty regarding collateral values and therefore poses an obstacle to liquidating collateral during debt enforcement. Difficulties in realizing collateral, in turn, increase banks’ reluctance to address NPLs. Requiring banks to have operational targets for NPL reduction or time limits on carrying NPLs on their balance sheets, where these tools are absent, can create incentives for resolving NPLs. Finally, as discussed earlier, provisioning ratios are low in the Caribbean compared with many other regions, suggesting scope for greater provisioning to help recognize losses and write off loans (Figure 11.13).

Ratio of Loan Loss Provisions to NPLs
Sources: IMF, Global Financial Stability Report, and national authorities.Note: Latest available data. CEE = Central and Eastern Europe; CIS = Commonwealth of Independent States; MENA = Middle East and North Africa; NPL = nonperforming loan.
Ratio of Loan Loss Provisions to NPLs
Sources: IMF, Global Financial Stability Report, and national authorities.Note: Latest available data. CEE = Central and Eastern Europe; CIS = Commonwealth of Independent States; MENA = Middle East and North Africa; NPL = nonperforming loan.Ratio of Loan Loss Provisions to NPLs
Sources: IMF, Global Financial Stability Report, and national authorities.Note: Latest available data. CEE = Central and Eastern Europe; CIS = Commonwealth of Independent States; MENA = Middle East and North Africa; NPL = nonperforming loan.Banks have the basic tools with which to address NPLs. Most banks have dedicated NPL workout units, and all banks are required to have NPL management strategies and action plans to reduce problem loans. In most countries, banks can also outsource NPL management. Survey responses suggest that banks use a variety of tools to restructure NPLs, including interest-only loans, reduced repayments, and performance-based write-offs (Figure 11.14). Restructured loans averaged about 10 percent of NPL portfolios in the sample, although they reached as high as 40 percent in a few banks. However, lack of coordination mechanisms among creditors and the absence of a market for problem loans constrain NPL resolution. Banks disposed of about 16 percent of their NPL portfolios (on average in 2013–15), mostly through write-offs. Difficulties liquidating collateral and uncertainties regarding collateral valuations create incentives to keep distressed assets on bank balance sheets much longer than warranted, despite the availability of resolution tools. NPL management is also constrained by the lack of interbank or public-private creditor coordination.

Use of Debt-Restructuring and NPL Disposal Tools
Source: IMF surveys of national authorities and banks.Note: AMC = asset management company; ECCU = Eastern Caribbean Currency Union; n.a. = not available; NPL = nonperforming loan.1 Compiled from supervisory authority responses in eight jurisdictions: ECCU, The Bahamas, Barbados, Belize, Guyana, Jamaica, Suriname, and Trinidad and Tobago.2 Compiled from responses of 35 banks from the following jurisdictions: ECCU, The Bahamas, Belize, Barbados, Guyana, Jamaica, Suriname, and Trinidad and Tobago.
Use of Debt-Restructuring and NPL Disposal Tools
Source: IMF surveys of national authorities and banks.Note: AMC = asset management company; ECCU = Eastern Caribbean Currency Union; n.a. = not available; NPL = nonperforming loan.1 Compiled from supervisory authority responses in eight jurisdictions: ECCU, The Bahamas, Barbados, Belize, Guyana, Jamaica, Suriname, and Trinidad and Tobago.2 Compiled from responses of 35 banks from the following jurisdictions: ECCU, The Bahamas, Belize, Barbados, Guyana, Jamaica, Suriname, and Trinidad and Tobago.Use of Debt-Restructuring and NPL Disposal Tools
Source: IMF surveys of national authorities and banks.Note: AMC = asset management company; ECCU = Eastern Caribbean Currency Union; n.a. = not available; NPL = nonperforming loan.1 Compiled from supervisory authority responses in eight jurisdictions: ECCU, The Bahamas, Barbados, Belize, Guyana, Jamaica, Suriname, and Trinidad and Tobago.2 Compiled from responses of 35 banks from the following jurisdictions: ECCU, The Bahamas, Belize, Barbados, Guyana, Jamaica, Suriname, and Trinidad and Tobago.Debt-Enforcement and Insolvency Framework
Effective insolvency regimes and debt enforcement are essential for NPL resolution. An effective insolvency regime should provide mechanisms for creditors to realize their claims in a predictable, speedy, and transparent manner, while at the same time protecting and maximizing value for all parties (Aiyar and others 2015). An effective debt-enforcement and insolvency framework has, in turn, two pillars: first, an adequate resolution toolkit that provides for rehabilitation for viable firms and liquidation for non-viable firms (and in the case of personal insolvency, a second chance for good faith entrepreneurs while preserving credit discipline), and second, an effective institutional framework that operates in a predictable, efficient, and transparent manner. Strengthening insolvency regimes by enhancing insolvency laws, creating incentives for out-of-court settlement, and institutional reforms such as creating specialized courts were key elements of strategies for reducing NPLs in a number of countries.12
The surveys highlighted several gaps in the resolution toolkit and its implementation. Even though all countries have corporate insolvency regimes in place, many do not have fast-track pre-pack insolvency procedures (where the court expeditiously approves a debt restructuring plan negotiated between the debtor and its creditors in a consensual manner before the initiation of an insolvency proceeding) or rehabilitation procedures, and only a few have out-of-court settlement mechanisms or actively promote such techniques (e.g., through guidance issued by the central bank or informal arrangements between banks). These gaps, in turn, impede business rescue, slow down realization of claims, and reduce expected recovery of value. Several countries reported that creditors do not have adequate control or influence over the process of business restructuring. Personal insolvency regimes exist in only half of the surveyed countries, and this regime is applicable to individual entrepreneurs in only two cases. Weak institutions are even a more significant concern—half of the surveyed banks consider implementation of insolvency and debt-enforcement regimes to be important obstacles to NPL resolution. Courts or judges that specialize in insolvency cases do not exist, and in the absence of time-bound procedures in many cases, resolving insolvency takes a long time, on average 2.7 years for the region—a high- or medium-level concern for about 60 percent of the bank respondents (Figure 11.15). In the ECCU region, for example, foreclosure laws are typically debtor-friendly, which lengthens considerably the foreclosure process and delays recovery by banks. These obstacles reduce recovery rates, with subsequent adverse effects on credit provision and lending rates.

Dealing with Insolvency: Time to Complete and the Cost of Insolvency
Source: World Bank, Doing Business Database.Note: Data labels in figure use International Organization for Standardization (ISO) country codes.
Dealing with Insolvency: Time to Complete and the Cost of Insolvency
Source: World Bank, Doing Business Database.Note: Data labels in figure use International Organization for Standardization (ISO) country codes.Dealing with Insolvency: Time to Complete and the Cost of Insolvency
Source: World Bank, Doing Business Database.Note: Data labels in figure use International Organization for Standardization (ISO) country codes.Data Gaps and Impediments to Information Sharing
Access to adequate information is essential for borrowers and creditors to price risk, value collateral, and narrow pricing gaps by reducing uncertainty. Public registries are important for maintaining complete information about credit and transaction histories and asset characteristics. The ability to share information for debt-workout purposes among creditors is also important. Reducing information asymmetries was a key pillar in strategies to reduce NPLs in many European economies.13
The surveys identified significant gaps in the information framework in the Caribbean (Figure 11.16). For example, 80 percent of banks consider that lack of credit registries significantly raises the cost of due diligence. Most countries do not have credit bureaus. Although most countries have land registries accessible to creditors, only four out of nine have registries of assets and real estate transactions, and access to the existing asset registries is more restricted. Even where registries exist, there are concerns about the quality of the data and access. For example, existing credit bureaus lack basic elements such as scoring for borrowers and information on connected borrowers. Four countries reported restrictions on recording and sharing of personal information for debt-workout purposes. These restrictions hinder the resolution of NPLs by undermining the debt-workout process in the absence of a full assessment of debtors’ liabilities and wealth. The importance of addressing these information gaps is highlighted by the experience of Jamaica, where the operationalization and increased use of credit bureaus since 2014 have helped incentivize borrowers to preserve good credit ratings and improve credit underwriting and management by deposit-taking institutions, and played a role in reducing new NPLs and recoveries.

Information Obstacles
(Percent of total responses)
Source: Survey of national authorities.Note: Based on responses from nine countries, n.a. = not available.
Information Obstacles
(Percent of total responses)
Source: Survey of national authorities.Note: Based on responses from nine countries, n.a. = not available.Information Obstacles
(Percent of total responses)
Source: Survey of national authorities.Note: Based on responses from nine countries, n.a. = not available.Market for NPLs
An active, liquid market for distressed debt can facilitate disposal of NPLs and reduce uncertainty about collateral valuations, thereby helping strengthen recovery values. Allowing banks to move problem loans off their balance sheets would reduce the burden on banks and can boost recovery values by providing a more cost-effective alternative to internal NPL management, especially for smaller banks that lack expertise and economies of scale in managing NPLs. Market approaches to dispose of NPLs may include direct sales, as well as securitization of NPLs. The latter has been used in a number of countries as a strategy for dealing with NPL overhang (e.g., Ireland and Spain) and as a useful approach to expanding the universe of distressed debt investors and to offering a way through which governments could jump-start the NPL market (e.g., by co-investing, together with private investors, in junior or mezzanine tranches) (see, for example, Aiyar and others 2015). Survey responses show that markets for NPLs in the Caribbean are largely nonexistent, including in countries that do not have specific restrictions on trading NPLs. Half of the banks consider the lack of markets to be a medium- or high-level concern and, hence, an obstacle to resolution. Most countries reported no prohibitions on sales of NPLs to domestic or foreign buyers (Figure 11.17).14 Yet NPL sales are either sporadic or nonexistent. Many banks point out that the lack of a market for NPLs stems from large pricing gaps: values offered by potential buyers are much lower than the prices at which banks would be willing to sell their distressed assets. This may be a particularly important concern for banks with low levels of provisioning and capital to absorb the losses that could arise from selling bad assets. These considerations are similar to those highlighted by the surveys of European authorities and banks in Aiyar and others (2015).

Restrictions in the Market for Distressed Assets
(Percent of total responses)
Source: Survey of national authorities.Note: AMC = asset management company; n.a. = not available; NPL = nonperforming loan. Based on responses from eight jurisdictions.
Restrictions in the Market for Distressed Assets
(Percent of total responses)
Source: Survey of national authorities.Note: AMC = asset management company; n.a. = not available; NPL = nonperforming loan. Based on responses from eight jurisdictions.Restrictions in the Market for Distressed Assets
(Percent of total responses)
Source: Survey of national authorities.Note: AMC = asset management company; n.a. = not available; NPL = nonperforming loan. Based on responses from eight jurisdictions.Tax Regime Obstacles
Unfavorable tax treatment can create disincentives for adequate provisioning and loan write-offs. The survey responses suggest that tax treatment of provisioning and write-offs are not a major concern for NPL resolution in the Caribbean (Figures 11.11 and 11.12). Only a small share of banks, most of which operate in two countries, flagged tax treatment as posing a medium or high concern for NPL resolution. This result stands in contrast with the actual rules, which suggest that tax treatment is not favorable in many countries. Half of the countries do not grant tax deductions for loan write-offs and loan loss provisioning, and two countries do not have a loss-carrying mechanism such as a deferred tax asset, thereby providing limited incentives for provisioning for or writing off impaired loans.
Way Forward: Strategy for NPL Resolution
The Caribbean NPL problem is a legacy of the global financial crisis, but its persistence at high levels owes much to continued weaknesses in economic growth in the region, as well as to structural impediments to their resolution. To address the problem, a multifaceted approach is needed involving macroeconomic policies to support growth and employment, prudential policies to ensure macro-financial stability, and a comprehensive strategy to reduce the structural bottlenecks for resolution of problem loans—including through tighter bank supervisory regimes; stronger insolvency; debt-enforcement; and legal frameworks; closing of information gaps; and development of markets for distressed debt (Figure 11.18) (see Liu and Rosenberg 2013; Bergthaler and others 2015; and Aiyar and others 2015).

Strategy to Address the Obstacles to NPL Resolution
Source: Authors’ illustration.Note: NPL = nonperforming loan.
Strategy to Address the Obstacles to NPL Resolution
Source: Authors’ illustration.Note: NPL = nonperforming loan.Strategy to Address the Obstacles to NPL Resolution
Source: Authors’ illustration.Note: NPL = nonperforming loan.Strong macroeconomic policies. With macroeconomic conditions identified as a key determinant of NPLs and as one of the main obstacles to their resolution, by banks and authorities in many countries, priority should be given to implementing macroeconomic policies and structural reforms that promote strong and sustainable growth. Evidence suggests that growth recoveries associated with lower unemployment, higher household incomes, and higher corporate profits help resolve problem loans by improving the ability of firms and households to service their loans.
Strengthening prudential and supervisory frameworks. Strengthening supervisory and prudential frameworks, where needed, would help prevent a buildup of problem loans in the future. While the region is still at a low growth stage of the credit cycle, authorities could preemptively set up a toolkit of well-targeted micro- and macroprudential policies to reduce excessive risk taking during loan origination and to mitigate emerging risks subsequently identified. Strong supervisory regimes are also key to addressing existing problem loans. Banks must follow prudent loss-recognition and provisioning practices and apply appropriate real estate valuation standards to improve collateral valuation and liquidation during debt enforcement. Incentives should be in place to write off bad loans and increase pro-visioning to facilitate recognition of losses.15 Policies should also focus on facilitating outsourcing of NPL management, including establishing a regulatory framework for nonbanks to manage NPLs and fostering bilateral NPL sales, undertaking independent asset quality reviews, as well as ensuring high quality of governance and senior management oversight in banks. Effective coordination mechanisms between banks and between public and private creditors should be in place.
Addressing informational obstacles. Policies should focus on removing information gaps that hinder the pricing of risk, collateral valuation, and the narrowing of large pricing gaps between buyers and sellers of properties that hamper development of NPL markets. Where missing, efforts could focus on establishing well-functioning public registries and credit bureaus to reduce information asymmetries and improve management of credit risk, and on setting up public records on assets or real estate transactions to improve the functioning and revival of real estate markets. The quality of data also needs improvement even when registries exist. Improving data quality and closing data gaps are crucial not only to facilitate NPL resolution and foster market development, but also to address other regional challenges, such as de-risking and assessing emerging risks in the financial sector.
Improving insolvency and debt-enforcement frameworks is essential for a predictable and transparent system that enables business rescue (for firms) or a second chance (for individuals) and helps creditors to realize claims promptly and protect recovery value. All countries have corporate bankruptcy regimes, but some countries could expand the toolkit by establishing fast-track and rehabilitation procedures and promoting out-of-court mechanisms for debt restructuring. Several countries could consider introducing personal bankruptcy mechanisms. Policies should focus on strengthening the institutions that support/implement the legal frameworks, for instance by improving judicial systems (a key obstacle noted in the surveys) to reduce the time required for resolution. Specialized courts and judges and experienced resolution professionals could strengthen the expertise necessary to facilitate timely and effective resolution.
Introducing a pan-Caribbean market for NPLs. The small size of the Caribbean economies, as well as structural obstacles such as gaps in the information framework and cultural and social factors, prevent development of a well-functioning domestic market for distressed assets to facilitate the disposal of NPLs. Access to timely information on distressed borrowers, collateral valuations, and recent NPL sales are critical for developing an active NPL market for distressed assets (Aiyar and others 2015). Policies should focus on building a market based on a common pan-Caribbean platform open to external investors to create economies of scale and address some of the cultural and social obstacles, while also ensuring adequate capital and provisioning levels. Introducing asset management companies (AMCs) can help initiate markets; the experience with AMCs in the region is limited but promising (Box 11.1).16 While NPL securitization may not be a viable solution for the Caribbean at this juncture, given the small and underdeveloped financial markets and weaknesses in legal and regulatory frameworks, efforts to develop the necessary ingredients for securitizing NPLs, including regulatory and legislative changes, could pave the way to using this tool as a means of pooling risk and facilitating the market for NPLs in the future.
While priority should be given to implementing the strategies outlined above, implementation may face challenges. Some of the strategy’s measures, such as enhanced supervision to ensure loss recognition and provision, can be introduced quickly. However, others, such as legal reforms, measures to address informational obstacles, and establishing a market for problem loans, require time to implement. The strategy needs to recognize that each country would require a separate diagnostic that could help in prioritizing and sequencing the reforms, given the differences within the region. The institutional capacity constraints in small Caribbean states call for coordination of reforms within the region. Some of the ongoing regional initiatives, such as establishing a regional AMC and credit bureaus, enhancing insolvency and debt-enforcement regimes, and establishing guidelines for collateral valuation in the Eastern Caribbean region, are positive steps in this regard.17 Reforms should also be fine-tuned based on the detailed institutional framework and coordinated across stakeholders, such as the central bank and ministries of finance. National and regional efforts could be supported by well-targeted capacity-building assistance from international financial organizations.
Asset Management Companies (AMCs) in the Caribbean
AMCs can enable more efficient collection and disposal of distressed assets and help maximize the value of impaired assets in the banking system, while at the same time preventing credit discipline from deteriorating. International evidence (see, for example, Klingebiel 2000) points to mixed experience in the use of AMCs and suggests that there is no single recipe for establishing a successful AMC. However, common factors for successful AMCs include a supporting legal and regulatory environment, strong leadership, operational independence, appropriate incentives, and a commercial orientation (Ingves, Seelig, and He 2004).
A large majority of Caribbean banks that responded to the IMF survey argue that banks can set up private AMCs and that indeed there are active AMCs operating in their countries. At the same time, there are few examples of operational public AMCs.
Eastern Caribbean Currency Union (ECCU). In the context of bank restructuring within the ECCU, the Eastern Caribbean Central Bank has created an AMC to resolve problem loans. A regional asset management company (ECAMC) has been granted extraordinary powers to acquire collateral and to resolve problem loans. ECAMC will acquire the problem loans from three now-defunct ECCU banks and will be able to purchase NPLs from other banks in the region. A new banking act was passed by all ECCU jurisdictions providing the legislation for the establishment of the ECAMC. It is anticipated that ECAMC will operate for three years.
The Bahamas. Both public and private AMCs have made some progress in removing bad loans from bank balance sheets. First, in 2014, the government established a special purpose vehicle (Bahamas Resolve) to manage $100 million of nonperforming commercial loan assets from the largely state-owned Bank of The Bahamas. There has reportedly been little to no progress in sale of the underlying real estate assets. Second, one commercial bank operating in The Bahamas sold a portfolio of NPLs (about $75 million) to a newly established specialized mortgage servicing company (Gateway). Anecdotal evidence suggests that the experience has been positive thus far, with no major consumer protection complaints related to loan restructuring and potential for other banks to follow suit.
Annex 11.1
Key Measures Implemented in the Past Three Years to Resolve NPLs
Key Measures Implemented in the Past Three Years to Resolve NPLs
Measures taken by country authorities | Measures taken by banks |
---|---|
|
|
Key Measures Implemented in the Past Three Years to Resolve NPLs
Measures taken by country authorities | Measures taken by banks |
---|---|
|
|



Panel VAR: Main Model (Impulse response functions, annual frequency)
Source: IMF staff estimates and calculations.Note: CPI = consumer price index; NPL = nonperforming loan; VAR = vector autoregression. Panel VAR includes change in NPL ratio, credit growth, change in unemployment rate, real GDP growth, CPI inflation. Shocks are of one standard deviation. Errors are 10 percent generated by Monte Carlo with 300 simulations. Red signifies the presence of a statistically significant effect.


Panel VAR: Main Model (Impulse response functions, annual frequency)
Source: IMF staff estimates and calculations.Note: CPI = consumer price index; NPL = nonperforming loan; VAR = vector autoregression. Panel VAR includes change in NPL ratio, credit growth, change in unemployment rate, real GDP growth, CPI inflation. Shocks are of one standard deviation. Errors are 10 percent generated by Monte Carlo with 300 simulations. Red signifies the presence of a statistically significant effect.


Panel VAR: Main Model (Impulse response functions, annual frequency)
Source: IMF staff estimates and calculations.Note: CPI = consumer price index; NPL = nonperforming loan; VAR = vector autoregression. Panel VAR includes change in NPL ratio, credit growth, change in unemployment rate, real GDP growth, CPI inflation. Shocks are of one standard deviation. Errors are 10 percent generated by Monte Carlo with 300 simulations. Red signifies the presence of a statistically significant effect.


Panel VAR: Main Model (Impulse response functions, annual frequency)
Source: IMF staff estimates and calculations.Note: CPI = consumer price index; NPL = nonperforming loan; VAR = vector autoregression. Panel VAR includes change in NPL ratio, credit growth, change in unemployment rate, real GDP growth, CPI inflation. Shocks are of one standard deviation. Errors are 10 percent generated by Monte Carlo with 300 simulations. Red signifies the presence of a statistically significant effect.


Panel VAR: Main Model (Impulse response functions, annual frequency)
Source: IMF staff estimates and calculations.Note: CPI = consumer price index; NPL = nonperforming loan; VAR = vector autoregression. Panel VAR includes change in NPL ratio, credit growth, change in unemployment rate, real GDP growth, CPI inflation. Shocks are of one standard deviation. Errors are 10 percent generated by Monte Carlo with 300 simulations. Red signifies the presence of a statistically significant effect.


Panel VAR: Main Model (Impulse response functions, annual frequency)
Source: IMF staff estimates and calculations.Note: CPI = consumer price index; NPL = nonperforming loan; VAR = vector autoregression. Panel VAR includes change in NPL ratio, credit growth, change in unemployment rate, real GDP growth, CPI inflation. Shocks are of one standard deviation. Errors are 10 percent generated by Monte Carlo with 300 simulations. Red signifies the presence of a statistically significant effect.Panel VAR: Main Model (Impulse response functions, annual frequency)
Source: IMF staff estimates and calculations.Note: CPI = consumer price index; NPL = nonperforming loan; VAR = vector autoregression. Panel VAR includes change in NPL ratio, credit growth, change in unemployment rate, real GDP growth, CPI inflation. Shocks are of one standard deviation. Errors are 10 percent generated by Monte Carlo with 300 simulations. Red signifies the presence of a statistically significant effect.References
Aiyar, S., W. Bergthaler, J. M. Garrido, A. Ilyina, A. Jobst, K. Kang, D. Kovtun, Y. Liu, D. Monaghan, and M. Moretti. 2015. “A Strategy for Resolving Europe’s Problem Loans.” IMF Staff Discussion Note 15/19, International Monetary Fund, Washington, DC.
Arellano, M., and S. Bond. 1991. “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations.” Review of Economic Studies 58 (2): 277–97.
Arellano, M., and O. Bover. 1995. “Another Look at the Instrumental Variable Estimation of Error-Components Models.” Journal of Econometrics 68 (1): 29—51.
Bank for International Settlements. 2016. Guidelines on Prudential Treatment of Problem Assets—Definitions of Nonperforming Exposures and Forbearance. Basel: Bank for International Settlements.
Beaton, K., T. Dowling, D. Kovtun, F. Loyola, A. Myrvoda, J. Okwuokei, I. Ötker, and J. Turunen. Forthcoming. “Problem Loans in the Caribbean: Determinants, Impact, and Strategies for Resolution.” IMF Working Paper, International Monetary Fund, Washington, DC.
Beaton, K., A. Myrvoda, and S. Thompson. 2016. “Non-Performing Loans in the ECCU: Determinants and Macroeconomic Impact.” IMF Working Paper 16/229, International Monetary Fund, Washington, DC.
Beck, R., P. Jakubίk, and A. Piloiu. 2013. “Non-Performing Loans: What Matters in Addition to the Economic Cycle?” European Central Bank Working Paper 1515, European Central Bank, Frankfurt.
Bergthaler, W., K. Kang, Y. Liu, and D. Monaghan. 2015. “Tackling Small and Medium Sized Enterprise Problem Loans in Europe.” IMF Staff Discussion Note 15/04, International Monetary Fund, Washington, DC.
Blundell, R., and S. Bond. 1998. “Initial Conditions and Moment Restrictions in Dynamic Panel Data Model.” Journal of Econometrics 87 (1): 115–43.
Constâncio, V. 2017. “Resolving Europe’s NPL Burden: Challenges and Benefits.” Keynote Speech “Tackling Europe’s Non-performing Loans Crisis: Restructuring Debt, Reviving Growth” organized by Bruegel, Brussels. February 2.
Espinoza, R., and A. Prasad. 2010. “Nonperforming Loans in the GCC Banking System and their Macroeconomic Effects.” IMF Working Paper 10/224, International Monetary Fund, Washington, DC.
Holtz-Eakin, D., W. Newey, and H. S. Rosen. 1988. “Estimating Vector Autoregressions with Panel Data.” Econometrica 56 (6): 1371–95.
Ingves, S., S. Seelig, and D. He. 2004. “Issues in the Establishment of Asset Management Companies.” IMF Policy Discussion Paper 04/3, International Monetary Fund, Washington, DC.
Jordan, A., and C. Tucker. 2013. “Assessing the Impact of Nonperforming Loans on Economic Growth in The Bahamas.” Monetaria I (2): 371–400.
Klein, N. 2013. “Non-Performing Loans in CESEE: Determinants and Impact on Macroeconomic Performance.” IMF Working Paper 13/72, International Monetary Fund, Washington, DC.
Klingebiel, D. 2000. “The Use of Asset Management Companies in the Resolution of Banking Crises: Cross-Country Experiences.” Policy Research Working Paper 2284, World Bank, Washington, DC.
Lexology. 2016. “Non-Performing Loans and Securitization in Europe.” https://www.lexology.com/library/detail.aspx?g=5907bb19-7e3e-4060-84c1-d73bcfaccc28.
Liu, Y., and R. Rosenberg. 2013. “Dealing with Private Debt Distress in the Wake of the European Financial Crisis: A Review of the Economics and Legal Toolbox.” IMF Working Paper 13/44, International Monetary Fund, Washington, DC.
Nickell, S. J. 1981. “Biases in Dynamic Models with Fixed Effects.” Econometrica 49 (6): 1417–1426.
Nkusu, M. 2011. “Nonperforming Loans and Macrofinancial Vulnerabilities in Advanced Economies.” IMF Working Paper 11/161, International Monetary Fund, Washington, DC.
Tintchev, K. Forthcoming. “Commodity Price Shocks and Bank Credit Quality in the Caribbean.” IMF Working Paper, International Monetary Fund, Washington, DC.
This chapter is a condensed version of a forthcoming IMF Working Paper (Beaton and others, forthcoming). It has benefited greatly from the contributions of Shelton Nicholls and from valuable comments and suggestions by Krishna Srinivasan, Dermot Monaghan, Diarmuid Murphy, and the authorities of The Bahamas, the ECCB, and Jamaica. The authors are grateful to the banks and country authorities who participated in this project through responses to the surveys on impediments to NPL resolution, as well as provision of bank-by-bank and country-level data for the empirical analyses.
The analysis in this chapter measures asset quality by the NPL ratio—a commonly used financial soundness indicator in cross-country comparisons, notwithstanding its well-known shortcomings. These shortcomings include (1) lack of uniformity in national definitions, (2) differences in treatment of restructured loans, and (3) its backward-looking nature. These caveats have prompted initiatives for further harmonization of definitions (for example, see Bank for International Settlements 2016).
These results are broadly consistent with those in Beaton, Myrvoda, and Tompson (2016) for the ECCU, Tintchev (forthcoming) for commodity-exporting Caribbean countries, Jordan and Tucker (2013) for The Bahamas, and similar studies for other regions (see, for example, Espinoza and Prasad 2010; Beck, Jakubίk, and Piloiu 2013; Klein 2013; Nkusu 2011).
In Jamaica, for example, the provisioning ratio rose sharply since 2011, exceeding 100 percent from 2014.
Estimates are produced using lagged regressors as instruments and estimated using generalized method of moments (Klein 2013). Macro-financial feedback effects are assessed using impulse response functions. Orthogonal shocks are identified using the Cholesky decomposition. In the baseline specification, NPLs appear first in the ordering, followed by credit growth, change in unemployment rate, real GDP growth, and CPI inflation.
The full sample includes 242 observations from the following countries: Antigua and Barbuda, The Bahamas, Barbados, Belize, Dominica, Grenada, Guyana, Jamaica, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, Suriname, and Trinidad and Tobago. Time series unemployment data are unavailable for Antigua and Barbuda, Dominica, Grenada, St. Kitts and Nevis, and St. Vincent and the Grenadines.
These include six members of the ECCU, The Bahamas, Barbados, Belize, Guyana, Jamaica, Suriname, and Trinidad and Tobago. For most countries, data are available beginning about 2000.
All models are estimated with dynamic panel regression techniques. To address dynamic panel bias (Nickell 1981) and endogeneity, all models are estimated with the system generalized method of moments (SGMM) developed in Arellano and Bover (1995) and Blundell and Bond (1998). The instrument set is restricted to avoid overfitting bias (see Arellano and Bond 1998). We use as instruments only the first and second appropriate lag of each explanatory variable. To further reduce the number of instruments, the collapsing method of Holtz-Eakin, Newey, and Rosen (1988) was used. In the model specifications, all global and country-specific variables, except for country-specific economic growth, are treated as strictly exogenous. All bank-specific variables and country-specific economic growth are treated as endogenous and instrumented with internal instruments (that is, lagged values of the instrumented variables).
An analysis of the determinants of NPLs across tourism-dependent and commodity-exporting countries suggests that the macroeconomic determinants of NPLs are broadly comparable for both types of Caribbean countries. See Beaton and others (forthcoming) for detailed results.
This is assessed by interacting a foreign bank dummy variable with all other determinants in the benchmark model. Detailed results are available in Beaton and others (forthcoming).
For most ECCU countries, Belize, and Suriname, domestic banks, on average, have higher NPL ratios compared with foreign banks—primarily subsidiaries and branches of international banks that benefit from stronger risk management practices and greater capacity to dispose of or write of impaired assets. On the other hand, in The Bahamas, Grenada, St. Vincent and the Grenadines, and Trinidad and Tobago, the NPL ratio is higher for foreign-owned banks on average.
The survey on the structural obstacles to NPL resolution was adapted from Aiyar and others (2015); questions were customized to ft the Caribbean region.
For example, in Indonesia and Thailand (1999), Turkey (2002), Japan (1999, 2008), and Korea (1998, 2006) (Aiyar and others 2015).
For example, Cyprus has set up a centralized credit registry, allowing banks to make more informed credit decisions. Serbia set up a database on real estate collateral valuations and loans secured by such collateral. Italy improved transparency and availability of NPL-specific data, fostering the development of a market for distressed assets.
One potential obstacle to market development and sale of NPLs is the Alien Land Holding License requirement in several Caribbean countries (for example, some ECCU countries). Regional efforts to remove these obstacles are in process, but countries must agree to waive the legislation to allow for cross-border land and property purchases, and there is some resistance to doing so.
Such incentives could include additional capital charges and time limits for carrying NPLs on balance sheets, tighter provisioning requirements, or tax incentives for write-offs and loan-loss provisioning.
AMCs (both private and public) have been used in some countries, particularly Asia, to facilitate NPL disposal (for example, in Indonesia, Korea, Malaysia, Sweden, and Thailand) by separating good assets from bad, allowing the ceding banks and AMCs to focus on their respective objectives, and by helping close the gap between the price at which banks were willing to sell and investors were willing to buy, and in so doing helping to nurture a market for distressed assets (Aiyar and others 2015).
These steps indeed form part of a broader comprehensive strategy by the ECCB to address the negative protracted impacts of the global financial crisis on the economies of the ECCU, strengthen the supervisory frameworks to ensure macro-financial stability, and create incentives for problem loan resolution and developing markets for distressed loans. Toward this goal, the ECCB is facilitating the implementation of an Eight-Point Programme approved by the Monetary Council in 2009, working with regional and international partners to build institutional capacity to bolster the resilience of the ECCU banking system (for example, by moving toward implementation of the Basel II framework, which is expected to help banks identify, measure, mitigate, and monitor their risks in a more proactive manner).